List your company

Region-wise comparison assessment on PANDABENCH (Hard set)

43AccuracyLinear Probe
10.7619.1327.535.87Apr 13, 2026

Evaluation results

MethodLinks
Linear ProbeType=Baseline2026.0443282114
PANDA2026.0440312524
Attentive ProbeType=Baseline2026.0439232118
DepictQA†Access=Open-source, Training=Trained on PANDASET2026.0433211819
GPT-5 NanoAccess=Closed-source2026.0421171616
RandomType=Baseline2026.0421202019
GPT-5 MiniAccess=Closed-source2026.0418192015
GPT-4oAccess=Closed-source2026.0416191614
Gemini 2.5 ProAccess=Closed-source2026.0412201412